Wolfram Computation Meets Knowledge

Wolfram Summer School


Carlos Muñoz

Science and Technology

Class of 2016


Carlos is currently studying biology as an undergraduate at the National Autonomous University of Mexico. He has been fascinated by insects his whole life, and currently is interested in studying collective behavior in eusocial insects like ants. Carlos has participated in projects related to conservation and the environment; one of them included a three-week journey in the arctic with the organization Cape Farewell to bring awareness about climate change. A year before entering university, he spent four months helping reintroduce the scarlet macaw and studying the behavior of the howler monkey in southern Mexico.

In the future, Carlos plans to continue his studies in the field of complexity applied to biological systems.

In his spare time, he breeds insects, learns to make bonsais and jewelry and plays video games.

Project: Anty McAntface, an Image Identifier for Ants

Anty McAntface is a neural network trained to identify images of ant faces to the level of genus. Currently it has an accuracy of 53% when trained with 80 images per class (with 73 classes—that’s 5,840 images in total). The first part consists of extracting features by passing photos of ant faces through the first 30 levels of Google’s Inception-v3 net. Then a new network is trained with the extracted features.

After that, to see the distribution of the subfamilies in the feature space, I reduced the dimensions of the features of each image to two and then plotted them by subfamily.

Favorite 3-Color 2D Totalistic Cellular Automaton

Rule 382938

I wanted to find a cellular automaton that had a behavior similar to that found in Conway’s Game of Life. To do that, I decided to make a state transition graph of the Game of Life, considering a 3*3 grid whose edges wrap around. I found that most initial conditions stayed like they were; the graph showed a lot of nodes that were only connected to themselves. But a large number of other nodes converged to one other node.

After searching for 3-color totalistic cellular automata with state transition graphs similar to the Game of Life, with a 2*2 grid, I found one that was somewhat similar. Then I tried this cellular automaton with a large random grid, and eventually found a moving structure, or a glider, that can move horizontally or vertically. With specific conditions, it can change its direction.